Financial Forecast: Systematic Analysis Using Principal Component Analysis

Abstract

The most popular technique for data exploration and analysis across all scientific disciplines that can influence the classification model's accuracy is principal component analysis (PCA). In this study, the high dimensionality of the stock market is examined in order to forecast market movements using PCA in conjunction with linear regression. While reducing data redundancy, PCA can aid in enhancing the predictive performance of machine learning techniques. The New York Stock Exchange, London Stock Exchange, and Karachi Stock Exchange are three stock exchanges that are the focus of experiments on a high-dimensional spectrum. Before and after using PCA, the accuracy of the linear regression classification model is evaluated. The results of the tests demonstrate that PCA can enhance machine learning performance in general, but only when the relative correlation between input features is studied and careful consideration is given while selecting principal components. The classification model is assessed using the root mean square error (RMSE) evaluation measure.